[R] "R is not a validated software package.."

Sat Jun 9 17:44:43 CEST 2007

You've just opened up another bit of confusion.   A submission package has 
many pieces, and that cited one is just a small part of it.

As Frank has mentioned (though perhaps tritely), and Cody points out -- 

The only issue that a Pharma has to worry about is whether they know enough 
about a software language/application to trust it.  Period.  There are 
incantations and approaches for this, but working through a corporation of 
diverse skills and knowledge, to justify the use of a particular tool 
(whether it is statistical, legal, chemical, etc) with in the corporation, 
brings up the need to follow whatever implementation the company has done 
using the guidelines from FDA, EMEA, PMDA, etc...

But it's the company guidelines, and the company's due diligence that is 
important.

NOT the external view.   So the idea of computer systems validation, and the 
qualification of programming languages such as R and SAS is a good one, since 
it helps you understand where the weaknesses in computed results might be, 
for weighting results for internal decision making and knowledge management.

But much of it has to be done internally.  The side problem of "few people 
using S in clinical statistics" is related to the risks that a company has in 
having people that know about R, etc.  That is the practical issue, solvable 
when budgets exist and when it falls into strategy.

Sorry for getting on the soap-box, but I've just had a nasty week dealing with 
the results from  simplified bullet points which skip the major important 
complexities which were handled badly by decisions made using the summaries.  
Ugly, ugly nonsense.

And Computer Systems validation is a canonical area where the above paragraph 
holds.

best,
-tony.

On Friday 08 June 2007, Sicotte, Hugues Ph.D. wrote:
> I may have overstated  things a bit.
>
> See section VIII
> http://www.fda.gov/CDER/GUIDANCE/2396dft.htm
>
> If you are analyzing data your statistical package does not necessarely
> have to be validated. You may have to show that the statistical methods
> are adequate/appropriate or that the results are reproduced with
> different softwares if you are using non-standard packages. By all
> tests, S-plus appears acceptable, do not know about R.
>
> However, If your statistical method is an intricate part of a test, then
> you do have to validate the system.
> This is becoming increasingly relevant for theragnostics.
>
> .. Which is why I said
> "Should they need to use those results in a report [where] that will
> matter to the FDA.."
> (I added the where .. It makes more sense)
>
>
>
> -----Original Message-----
> From: Frank E Harrell Jr [mailto:f.harrell at vanderbilt.edu]
> Sent: Friday, June 08, 2007 11:08 AM
> To: Sicotte, Hugues Ph.D.
> Cc: Wensui Liu; Giovanni Parrinello; r-help at stat.math.ethz.ch
> Subject: Re: [R] "R is not a validated software package.."
>
> Sicotte, Hugues Ph.D. wrote:
> > People, don't get angry at the pharma statistician, he is just trying
>
> to
>
> > abide by an FDA requirement that is designed to insure that test
>
> perform
>
> > reliably the same. There is no point in getting into which product is
> > better. As far as the FDA rules are concerned a validated system beats
>
> a
>
> > "better" system any day of the week.
>
> There is no such requirement.
>
> > Here is your polite answer.
> > You can develop and try your software in R.
> > Should they need to use those results in a report that will matter to
> > the FDA, then you can work together with him to set up a validated
> > environment for S-plus. You then have to commit to port your code to
> > S-plus.
>
> That doesn't follow.  What matters is good statistical analysis practice
>
> no matter which environment you use.  Note that more errors are made in
> the data preparation / derived variables stage than are made by
> statistical software.
>
> Frank
>
> > As I assume that you do not work in a regulated environment, you
> > probably wouldn't have access to a validated SAS environment anyways.
>
> It
>
> > is not usually enough to install a piece of software, you have to
> > validate every step of the installation. Since AFAIK the FDA uses
> > S-plus, it would be to your pharma person's advantage to speed-up
> > submissions if they also had a validated S-plus environment.
>
> http://www.msmiami.com/custom/downloads/S-PLUSValidationdatasheet_Final.
>
> > pdf
> >
> >
> > -----Original Message-----
> > From: r-help-bounces at stat.math.ethz.ch
> > [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Wensui Liu
> > Sent: Friday, June 08, 2007 9:24 AM
> > To: Giovanni Parrinello
> > Cc: r-help at stat.math.ethz.ch
> > Subject: Re: [R] "R is not a validated software package.."
> >
> > I like to know the answer as well.
> > To be honest, I really have hard time to understand the mentality of
> > clinical trial guys and rather believe it is something related to job
> > security.
> >
> > On 6/8/07, Giovanni Parrinello <parrinel at med.unibs.it> wrote:
> >> Dear All,
> >> discussing with a statistician of a pharmaceutical company I received
> >> this answer about the statistical package that I have planned to use:
> >>
> >> As R is not a validated software package, we would like to ask if it
> >> would rather be possible for you to use SAS, SPSS or another approved
> >> statistical software system.
> >>
> >> Could someone suggest me a 'polite' answer?
> >> TIA
> >> Giovanni
> >>
> >> --
> >> dr. Giovanni Parrinello
> >> External Lecturer
> >> Medical Statistics Unit
> >> Department of Biomedical Sciences
> >> Viale Europa, 11 - 25123 Brescia Italy
> >> Tel: +390303717528
> >> Fax: +390303717488
> >> email: parrinel at med.unibs.it
> >>
> >>
> >>         [[alternative HTML version deleted]]
> >>
> >> ______________________________________________
> >> R-help at stat.math.ethz.ch mailing list
> >> https://stat.ethz.ch/mailman/listinfo/r-help
> >> PLEASE do read the posting guide
> >
> > http://www.R-project.org/posting-guide.html
> >
> >> and provide commented, minimal, self-contained, reproducible code.

-- 
best,
-tony

blindglobe at gmail.com
Muttenz, Switzerland.
"Commit early,commit often, and commit in a repository from which we can 
easily
roll-back your mistakes" (AJR, 4Jan05).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url : https://stat.ethz.ch/pipermail/r-help/attachments/20070609/6fb82306/attachment.bin 